DNA Methylation Patterns and Epigenetic Memory

Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

REVIEW

DNA methylation patterns and epigenetic memory

Adrian Bird1 Wellcome Trust Centre for Cell Biology, University of Edinburgh, Edinburgh EH9 3JR, UK

The character of a cell is defined by its constituent pro- (Pc-G/trx) protein complexes. (Histone modification has teins, which are the result of specific patterns of gene some attributes of an epigenetic process, but the issue of expression. Crucial determinants of gene expression pat- heritability has yet to be resolved.) This review concerns terns are DNA-binding transcription factors that choose DNA methylation, focusing on the generation, inheri- genes for transcriptional activation or repression by rec- tance, and biological significance of genomic methyl- ognizing the sequence of DNA bases in their promoter ation patterns in the development of mammals. Data regions. Interaction of these factors with their cognate will be discussed favoring the notion that DNA methyl- sequences triggers a chain of events, often involving ation may only affect genes that are already silenced by changes in the structure of chromatin, that leads to the other mechanisms in the embryo. Embryonic transcrip- assembly of an active transcription complex (e.g., Cosma tion, on the other hand, may cause the exclusion of the et al. 1999). But the types of transcription factors present DNA methylation machinery. The heritability of meth- in a cell are not alone sufficient to define its spectrum of ylation states and the secondary nature of the decision to gene activity, as the transcriptional potential of a ge- invite or exclude methylation support the idea that DNA nome can become restricted in a stable manner during methylation is adapted for a specific cellular memory development. The constraints imposed by developmen- function in development. Indeed, the possibility will be tal history probably account for the very low efficiency discussed that DNA methylation and Pc-G/trx may rep- of cloning animals from the nuclei of differentiated cells resent alternative systems of epigenetic memory that (Rideout et al. 2001; Wakayama and Yanagimachi 2001). have been interchanged over evolutionary time. Animal A “transcription factors only” model would predict that DNA methylation has been the subject of several recent the gene expression pattern of a differentiated nucleus reviews (Bird and Wolffe 1999; Bestor 2000; Hsieh 2000; would be completely reversible upon exposure to a new Costello and Plass 2001; Jones and Takai 2001). For re- spectrum of factors. Although many aspects of expres- cent reviews of plant and fungal DNA methylation, see sion can be reprogrammed in this way (Gurdon 1999), Finnegan et al. (2000), Martienssen and Colot (2001), and some marks of differentiation are evidently so stable that Matzke et al. (2001). immersion in an alien cytoplasm cannot erase the memory. Variable patterns of DNA methylation in animals The genomic sequence of a differentiated cell is thought to be identical in most cases to that of the zy- A prerequisite for understanding the function of DNA gote from which it is descended (mammalian B and T methylation is knowledge of its distribution in the ge- cells being an obvious exception). This means that the nome. In animals, the spectrum of methylation levels marks of developmental history are unlikely to be and patterns is very broad. At the low extreme is the caused by widespread somatic mutation. Processes less nematode worm Caenorhabditis elegans, whose genome irrevocable than mutation fall under the umbrella term lacks detectable m5C and does not encode a conven- “epigenetic” mechanisms. A current definition of epige- tional DNA methyltransferase. Another invertebrate, netics is: “The study of mitotically and/or meiotically the insect Drosophila melanogaster, long thought to be heritable changes in gene function that cannot be ex- devoid of methylation, has a DNA methyltransferase- plained by changes in DNA sequence” (Russo et al. like gene (Hung et al. 1999; Tweedie et al. 1999) and is 1996). There are two epigenetic systems that affect ani- reported to contain very low m5C levels (Gowher et al. mal development and fulfill the criterion of heritability: 2000; Lyko et al. 2000), although mostly in the CpT di- DNA methylation and the Polycomb-trithorax group nucleotide rather than in CpG, which is the major target for methylation in animals. Most other invertebrate ge- nomes have moderately high levels of methyl-CpG con- centrated in large domains of methylated DNA separated by equivalent domains of unmethylated DNA (Bird et al. 1E-MAIL [email protected]; FAX 0131-650-5379. Article and publication are at http://www.genesdev.org/cgi/doi/10.1101/ 1979; Tweedie et al. 1997). This mosaic methylation pat- gad.947102. tern has been confirmed at higher resolution in the sea

6 GENES & DEVELOPMENT 16:6–21 © 2002 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/02 $5.00; www.genesdev.org Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

DNA methylation and epigenetic memory squirt, Ciona intestinalis (Simmen et al. 1999). At the events occur in germ cells or the early embryo (Jaenisch opposite extreme from C. elegans are the vertebrate ge- et al. 1982), suggesting that de novo methylation is par- nomes, which have the highest levels of m5C found in ticularly active at these stages. There is evidence, how- the animal kingdom. Vertebrate methylation is dis- ever, that de novo methylation can also occur in adult persed over much of the genome, a pattern referred to as somatic cells. A significant fraction of all human CpG global methylation. The variety of animal DNA methyl- islands are prone to progressive methylation in certain ation patterns highlights the possibility that different tissues during aging (for review, see Issa 2000), or in ab- distributions reflect different functions for the DNA normal cells such as cancers (for review, see Baylin and methylation system (Colot and Rossignol 1999). Herman 2000) and permanent cell lines (Harris 1982; An- tequera et al. 1990; Jones et al. 1990). The rate of accu- mulation of methylated CpGs in somatic cells appears to Mammalian DNA methylation patterns vary in time be very slow. For example, de novo methylation of a and space provirus in murine erythroleukemia cells took many In human somatic cells, m5C accounts for ∼1% of total weeks to complete (Lorincz et al. 2000). Similarly, the DNA bases and therefore affects 70%–80% of all CpG recovery of global DNA methylation levels following dinucleotides in the genome (Ehrlich 1982). This average chronic treatment of mouse cells with the DNA meth- pattern conceals intriguing temporal and spatial varia- ylation inhibitor 5-azacytidine required months (Flatau tion. During a discrete phase of early mouse develop- et al. 1984). ment, methylation levels in the mouse decline sharply How do patterns of methylated and unmethylated to ∼30% of the typical somatic level (Monk et al. 1987; mammalian DNA arise in development and how are Kafri et al. 1992). De novo methylation restores normal they maintained? Why are CpG islands usually, but not levels by the time of implantation. A much more limited always, methylation-free? What causes methylation of drop in methylation occurs in the frog Xenopus laevis bulk non-CpG-island DNA? These burning questions (Stancheva and Meehan 2000), and no drop is seen in the cannot be answered definitively at present, but there are zebrafish, Danio rerio (MacLeod et al. 1999). Even within distinct hypotheses that have been addressed experimen- vertebrates, therefore, interspecies variation is seen that tally. The available data will be conveniently considered could reflect differences in the precise role played by in three parts: (1) mechanisms for maintaining DNA methylation in these organisms. For mice and probably methylation patterns; (2) mechanisms and consequences other mammals, however, the cycle of early embryonic of methylation gain; and (3) mechanisms and conse- demethylation followed by de novo methylation is criti- quences of methylation loss. cal in determining somatic DNA methylation patterns. A genome-wide reduction in methylation is also seen in Maintenance methylation—not so simple primordial germ cells (Tada et al. 1997; Reik et al. 2001) during the proliferative oogonial and spermatogonial Maintenance methylation describes the processes that stages. reproduce DNA methylation patterns between cell gen- The most striking feature of vertebrate DNA methyl- erations. The simplest conceivable mechanism for main- ation patterns is the presence of CpG islands, that is, tenance depends on semiconservative copying of the pa- unmethylated GC-rich regions that possess high relative rental-strand methylation pattern onto the progeny densities of CpG and are positioned at the 5Ј ends of DNA strand (Holliday and Pugh 1975; Riggs 1975). In many human genes (for review, see Bird 1987). Compu- keeping with the model, the methylating enzyme tational analysis of the human genome sequence pre- DNMT1 prefers to methylate those new CpGs whose dicts 29,000 CpG islands (Lander et al. 2001; Venter et al. partners on the parental strand already carry a methyl 2001). Earlier studies estimated that ∼60% of human group (Bestor 1992; Pradhan et al. 1999). Thus a pattern genes are associated with CpG islands, of which the of methylated and nonmethylated CpGs along a DNA great majority are unmethylated at all stages of develop- strand tends to be copied, and this provides a way of ment and in all tissue types (Antequera and Bird 1993). passing epigenetic information between cell generations. Because many CpG islands are located at genes that have The idea that mammalian DNA methylation patterns a tissue-restricted expression pattern, it follows that are established in early development by de novo meth- CpG islands can remain methylation-free even when yltransferases DNMT3A and DNMT3B (Okano et al. their associated gene is silent. For example, the tissue- 1998a, 1999; Hsieh 1999b) and then copied to somatic specifically expressed human ␣-globin (Bird et al. 1987) cells by the maintenance DNA methyltransferase and ␣ 2(1) collagen (McKeon et al. 1982) genes have CpG DNMT1 is elegant and simple, but, as discussed below, islands that remain unmethylated in all tested tissues, may not fully explain persistence of methylation pat- regardless of expression. terns during cell proliferation. A small but significant proportion of all CpG islands Experiments that first showed replication of methyl- become methylated during development, and when this ation patterns on artificially methylated DNA also re- happens the associated promoter is stably silent. Devel- vealed a relatively low fidelity for the process (Pollack et opmentally programmed CpG-island methylation of this al. 1980; Wigler et al. 1981). After many cell generations, kind is involved in genomic imprinting and X chromo- methylation of the introduced DNA was retained, but at some inactivation (see below). The de novo methylation a much lower level than in the starting plasmid. The

GENES & DEVELOPMENT 7 Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

Bird failure of maintenance was estimated to occur with a methyltransferases. DNMT3B in particular is known to frequency of ∼5% per CpG site per cell division. Quan- be required for de novo methylation of specific genomic titative studies of an endogenous CpG site broadly regions, as mice or human patients with DNMT3B mu- agreed with this figure (Riggs et al. 1998). Cell clones in tations are deficient in methylation of pericentromeric which this site was initially unmethylated acquired repetitive DNA sequences and at CpG islands on the methylation and clones where it was methylated lost inactive X chromosome (Miniou et al. 1994; Okano et al. methylation. The rate of change was estimated at ∼4% 1998b; Hansen et al. 2000; Kondo et al. 2000). DNMT3B per cell generation. Error rates of this magnitude mean may therefore be adapted to methylate regions of silent that a detailed methylation pattern would eventually be- chromatin. come indistinct as cells proliferate. Indeed, dynamic Evidence that accessory factors are also needed to en- changes in detailed methylation patterns have been ob- sure appropriate methylation came initially from plants, served in monoclonal lyomyomas (Silva et al. 1993) and where the SNF2-like protein DDM1 was shown to be at the methylated FMR1 gene (Stöger et al. 1997). These essential for full methylation of the Arabidopsis studies established that clonal populations of cells do thaliana genome (Jeddeloh et al. 1999). An equivalent not have the homogeneous methylation patterns that dependence is seen in animals, as mutations in human would be predicted by the replication model of mainte- ATRX (Gibbons et al. 2000) and mouse Lsh2 genes (Den- nance methylation. Not only does DNA methyltransfer- nis et al. 2001), both of which encode relatives of the ase fail to complete half-methylated sites at a significant chromatin-remodeling protein SNF2, have significant ef- rate, but also significant de novo methylation occurs at fects on global DNA methylation patterns. Loss of LSH2 unmethylated sites. protein, in particular, matches the phenotype of the At first sight, these findings appear to undermine the DDM1 mutation in Arabidopsis, for both mutants lose concept of maintenance methylation, but this does not methylation of highly repetitive DNA sequences, but re- follow. Although detailed methylation patterns may not tain some methylation elsewhere in the genome. Per- be maintained at the level of a single CpG nucleotide, haps efficient global methylation of the genome requires the methylation status of DNA domains appears to be perturbation of chromatin structure by these chromatin- faithfully propagated during development (Pfeifer et al. remodeling proteins so that DNMTs can gain access to 1990). CpG islands, for example, keep their overall un- the DNA. Collaboration between DNMTs and factors methylated state (or methylated state) extremely stably that allow them access to specialized chromosomal re- through multiple cell generations. DNMT1 is partly re- gions may be particularly important in regions that are sponsible for this stability, but there is likely to be an- heterochromatic and inaccessible. Although the net re- other as yet unknown component to the maintenance sult of these processes is apparently global genomic process. Dramatic evidence for this alternative mainte- methylation, the evidence for selectivity means that the nance mechanism comes from the finding that CpG-is- word “default” is probably not appropriate. land methylation is stably maintained even in the appar- ent absence of the only known maintenance DNA meth- Targeting de novo methylation to preferred yltransferase, DNMT1 (Rhee et al. 2000). A similar DNA sequences phenomenon may account for the maintenance of allele- specific DNA methylation imprints under conditions Another hypothesis to explain global methylation is that where the concentration of DNMT1 is severely limiting the DNA methylation machinery is preferentially at- (Jaenisch 1997). tracted by certain DNA sequences in the mammalian genome (Turker 1999). The presence of high levels of methylation in DNA outside such a DNA methylation De novo DNA methylation by default? center could be explained by spreading into the sur- The origin of DNA methylation patterns is a long-stand- rounding DNA. Barriers to spreading would lead to the ing mystery in the field. The de novo methyltransferases formation of CpG islands. A hypothetical trigger for DNMT3A and DNMT3B (Okano et al. 1998a, 1999) are DNA methylation is DNA sequence repetition, which highly expressed in early embryonic cells, and it is at this can promote de novo methylation in filamentous fungi stage that most programmed de novo methylation events and plants under certain circumstances (Selker 1999; occur. What determines which regions of the genome Martienssen and Colot 2001). The most suggestive evi- should be methylated? An extreme possibility is that de dence in mammals concerns manipulation of transgene novo DNA methylation in early mammalian develop- copy number at a single locus in the mouse genome us- ment is an indiscriminate process potentially affecting ing cre-lox technology (Garrick et al. 1998). High levels all CpGs. Compatible with the default model is the ap- of transgene repetition were found to cause significant parent absence of intrinsically unmethylatable DNA se- transgene silencing and concomitant methylation. The quences in mammalian genomes. Even CpG islands, efficiency of expression increased as copy number was most of which are unmethylated at all times in normal reduced at the locus, and the level of methylation de- cells, can acquire methylation under special develop- creased. Whether repetition caused methylation directly, mental circumstances or in abnormal cells (permanent or indirectly as a consequence of some other event (e.g., cell lines or cancer cells). It is clear, however, that not all transcriptional silencing; see below), is not known. regions of the genome are equally accessible to DNA The clearest definition of a DNA methylation center

8 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

DNA methylation and epigenetic memory comes from the fungus Neurospora, where short TpA- of posttranscriptional gene silencing in plants. Double- rich segments of DNA were found to induce methylation stranded RNA directs the destruction of transcripts con- (Miao et al. 2000). Identification of a mammalian DNA taining the same sequence, but there is compelling evi- methylation center located upstream of the mouse ad- dence that it can also direct de novo methylation of ho- enine phophoribosyltransferase (APRT) gene has been re- mologous genomic DNA (Wassenegger et al. 1994; ported (Mummaneni et al. 1993; Yates et al. 1999). The Bender 2001; Matzke et al. 2001). Posttranscriptional region contains B1 repetitive elements and attracts high gene silencing by double-stranded RNA is probably an levels of de novo methylation upon transfection into em- ancient genome defence system because it occurs in bryonic cells, although the effect is relative, because fungi, plants, and animals; but DNA methylation is not many DNA sequences are subject to de novo methyl- an obligatory accompaniment, as silencing is efficient in ation in these cells. The APRT methylation center be- C. elegans in the complete absence of genomic m5C. comes methylated in DNMT1-deficient ES cells, sup- Even in the fungus Neurospora, where transgene arrays porting the idea that it corresponds to a region that is a are often methylated, DNA methylation is not required favorable substrate for de novo methylation (Yates et al. for posttranscriptional gene silencing (or quelling; Co- 1999). goni et al. 1996). There are also specific features of RNA- Because the evidence suggests that replication of directed DNA methylation that may not occur in ani- methylation patterns by DNMT1 is only partly respon- mals; notably the occurrence of methylation at multiple sible for maintenance methylation (see above), an attrac- non-CpG cytosines in an affected DNA sequence tract. tive possibility is that the features of a DNA domain that Although there is evidence for non-CpG methylation in help maintain its methylated status are the same fea- ES cells, most probably owing to DNMT3A, which tures that promote its de novo methylation. Imprinting strongly methylates CpA as well as CpG (Ramsahoye et boxes, for example, whose differential methylation is as- al. 2000; Gowher and Jeltsch 2001), non-CpG methyl- sociated with genomic imprinting (Tremblay et al. 1997; ation is barely detectable in adult cells (Ramsahoye et al. Birger et al. 1999; Shemer et al. 2000), tend to retain their 2000). Plants have a CpG methylation system, but it methylation levels tenaciously even when the amount of does not appear to be essential for RNA-directed gene the maintenance enzyme DNMT1 is reduced (Beard et silencing (for reviews, see Wassenegger et al. 1994; al. 1995). The de novo methylases DNMT3A and Bender 2001; Matzke et al. 2001). Optimism that RNA- DNMT3B (Okano et al. 1998a, 1999) may be attracted directed de novo methylation will also apply in mam- disproportionately to these sequences, and this attrac- mals is tempered by this sequence disparity, and by the tion may also underlie the decision to methylate the box absence so far of a clear demonstration that mammalian in the first place. In other words, de novo methylation double-stranded RNA leads to DNA methylation-medi- may not occur once at a discrete and perhaps rather in- ated gene silencing. accessible stage of germ-cell development, but may hap- pen repeatedly (assisted by DNMT1) as embryonic cells Transcriptionally silent chromatin as a de novo divide. methylation target Several lines of evidence suggest that DNA methylation Unusual DNA structures and RNAi as triggers does not intervene to silence active promoters, but af- for de novo methylation fects genes that are already silent. It was reported many Studies of purified DNMT1 revealed that the enzyme years ago that retroviral transcription is repressed in em- prefers to methylate unusual DNA structures in vitro bryonic cells at ∼2 d after infection, whereas de novo (Smith et al. 1991; Laayoun and Smith 1995). This led to methylation is delayed until ∼15 d (Gautsch and Wilson the idea that such structures might be generated during 1983; Niwa et al. 1983). De novo methylation of proviral recombination between repetitive elements or during sequences in embryo cells depends on DNMT3A and transposition events and directly trigger de novo meth- DNMT3B (Okano et al. 1999), but initial retroviral shut- ylation (Bestor and Tycko 1996). Subsequent evidence, down occurs as usual even when both these de novo however, does not support a role for DNMT1 in de novo methyltransferases are absent (Pannell et al. 2000). methylation in vivo (Lyko et al. 1999; Howell et al. Clearly, de novo methylation is not required for silenc- 2001), and therefore the biological significance of its ing in the first instance, reinforcing the view that meth- predilection for deformed DNA is uncertain. There is ylation is a secondary event. evidence for transfer of methylation from one copy of a Methylation of genes that are already silent is also sequence to a second previously unmethylated copy of observed during X chromosome inactivation in the the same sequence in the fungus Ascobolus (Colot et al. mammalian embryo. Kinetic studies showed that the 1996). The process might use mechanisms involved in phosphoglycerate kinase gene is silent on the mamma- homologous DNA recombination and may therefore in- lian inactive X chromosome before methylation of its volve deformation of DNA. How identical sequences CpG-island promoters occurs (Lock et al. 1987). Subse- sense one another and transfer epigenetic information quent studies of the mouse, in which the process is best remains unknown, however. understood, have established that expression of a non- Exciting recent developments in the DNA methyl- coding chromosomal RNA from the Xist gene on the ation field have arisen through molecular genetic studies inactive X chromosome triggers the inactivation process

GENES & DEVELOPMENT 9 Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

Bird in cis. Specifically, activation of the Xist gene and onset al. 2000; Bachman et al. 2001) and the identification of of its late replication precede CpG-island methylation by genes that modify DNA methylation patterns (Weng et several days (Keohane et al. 1996; Wutz and Jaenisch al. 1995). 2000). In other words, methylation affects the X chromo- some on which genes are already shut down by other Consequences of methylation gain: stable mechanisms. Is transcriptional inertia during embryo- transcriptional silencing of genes genesis the trigger for de novo methylation? Studies of the origin of methylation-free CpG islands offer some Why methylate genes that are already silent? A plausible support for this idea. The coincidence between CpG is- answer is: to silence them irrevocably. Methylation lands and promoters is striking (Bird 1987), and foot- clearly contributes to the stability of inactivation, be- printing shows that the 5Ј extremity of CpG islands of- cause both X inactivation (Mohandas et al. 1981a; ten corresponds to the region occupied by transcription Graves 1982; Venolia et al. 1982) and retroviral silencing factors in vivo (Cuadrado et al. 2001). Even when CpG (Stewart et al. 1982; Jaenisch et al. 1985) can be relieved islands are identified in unusual locations, they have by treatment of somatic cells with demethylating turned out to correspond to promoters. For example, a agents. Individuals who lack DNMT3B show reduced CpG island located in intron 2 of the Igf2r gene is an methylation of some CpG islands on the inactive X chro- active promoter (Wutz et al. 1997; Lyle et al. 2000), as is mosome and also silence X-linked genes imperfectly a CpG island that covers exon 2 of the class II major (Miniou et al. 1994; Hansen et al. 2000). The implication histocompatibility gene (MacLeod et al. 1998). The po- that irreversibility involves DNA methylation is sup- tential importance of promoter function in the genesis of ported by the frequent reactivation of an X-linked trans- CpG islands is highlighted by studies in transgenic mice. gene in mouse embryo cells and in cultured somatic cells CpG-island-containing transgenes normally faithful re- when DNMT1 is absent or inhibited (Sado et al. 2000). produce their methylation-free character, but their im- This view is sustained by differences in the stability of munity to methylation is lost if promoter function is inactivity states pre- and postmethylation. For example, impaired (Brandeis et al. 1994; MacLeod et al. 1994). X inactivation caused by expression of an Xist transgene Similarly, viral DNA integrated into ES cell genomes by in embryonic stem cells is initially reversed when the homologous recombination becomes methylated when Xist gene is shut down, but after 3 d, inactivation be- the promoter is weakened by absence of an enhancer, but comes irreversible and independent of Xist (Wutz and excludes methylation when an enhancer is present Jaenisch 2000). Irreversibility may reflect the arrival of (Hertz et al. 1999). A parsimonious interpretation of the promoter methylation. results is that failure to transcribe invites de novo meth- In artificial systems, DNA methylation represses tran- ylation (see Fig. 2 below), although other potential ex- scription in a manner that depends on the location and planations (Brandeis et al. 1994; Mummaneni et al. 1998) density of the methyl-CpGs relative to the promoter cannot be discounted. (Boyes and Bird 1992; Hsieh 1994; Kass et al. 1997a,b). The signal for this putative gene silence-related de But what genes are affected by DNA methylation-medi- novo methylation is unknown, but the possibility that ated gene silencing? Early studies relied on the use of the chromatin states inform the DNA methylation machin- demethylating drug 5-azacytidine (Jones and Taylor ery is attractive (Selker 1990). The acetylation and 1980), which was shown to activate genes on the inac- methylation state of nucleosomal histones is tightly cor- tive X in rodent–human cell hybrids (Mohandas et al. related with transcriptional activity (Jenuwein and 1981b; Graves 1982). More recently, mice and murine Allis 2001) and could be read by the methylation ma- cell lines lacking DNMT1 (Li et al. 1992) have clarified chinery, leading it to either methylate or fail to methyl- the effects of DNA methylation on gene expression. In ate a particular domain. Indeed, recent work on Neuros- placental mammals, repression of X-linked genes fol- pora (Tamaru and Selker 2001) has shown an intimate lows expression of Xist, which sets in train the inactiva- link between histone methylation and DNA methyl- tion process, culminating in widespread methylation of ation in that fungus, as mutation of a histone methyl- CpG islands. The active X chromosome, on the other transferase that methylates Lys 9 of histone H3 abol- hand, must be protected from silencing, and this requires ished genomic methylation. In mammalian and yeast repression of Xist and again depends on methylation systems, histone H3 Lys 9 methylation is associated (Panning and Jaenisch 1996). An intact DNA methyl- with transcriptionally repressed heterochromatin (Ban- ation system is also essential for genomic imprinting, nister et al. 2001; Nakayama et al. 2001; Noma et al. because deletion of Dnmt1 leads to disruption of the 2001; Zhang and Reinberg 2001). If the dependence of monoallelic expression of several imprinted genes (Li et DNA methylation on prior histone methylation turns al. 1993). out to be applicable to mammals, this would further Both X inactivation and genomic imprinting involve strengthen the argument that DNA methylation is tar- silencing of one allele only, leaving the other unaffected. geted to genes that are already silent. The nature of the An unusual set of genes that are active in the germ line, molecular cues that trigger transfer of methyl groups to most of which are X-linked, appears to use methylation unmethylated DNA should be illuminated by ongoing for complete silencing in somatic cells (De Smet et al. studies of multiprotein complexes that contain DNA 1996, 1999). Several of the human and murine MAGE methyltransferases (Fuks et al. 2000, 2001; Robertson et genes, for example, have CpG-island promoters that are

10 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

DNA methylation and epigenetic memory methylation-free in germ cells, but are methylated in methylation levels in the hybrid embryo, the emergent somatic cells of the adult. The genes were discovered as element became transpositionally hyperactive, being tar- novel antigens in tumors, where genomic methylation geted exclusively to one parental genome. The parents of levels are often low and MAGE-gene CpG islands are the hybrid were not available to verify this unprec- undermethylated. MAGE expression can be induced by edented scenario. treating nonexpressing cells with demethylating agents, Phylogenetic studies of genomic methylation patterns supporting the idea that methylation is an important in animals have not yet offered support for the genome component of the repression of these genes in somatic defence model. Effective silencing due to sequence rep- cells. etition has been observed in Drosophila and C. elegans, but it is associated with the polycomb group of proteins or posttranscriptional gene silencing (Birchler et al. Transposable element silencing as a consequence 2000). The possibility that the low level of m5CinDro- of DNA methylation sophila (Lyko et al. 2000) is relevant to silencing has not Another well-documented consequence of DNA meth- yet been addressed. Studies of the sea squirt C. intesti- ylation deficiency is the activation of transposable ele- nalis, a chordate belonging to the same phylum as ver- ment-derived promoters. Like much of the mammalian tebrates, but which does not exhibit global methylation genome, transposable element-related sequences are of the genome, revealed that genes were often present in heavily methylated and transcriptionally silent in so- domains of methylated DNA, whereas transposable ele- matic cells. Mouse cells, for example, normally repress ment families, some of which appeared to be mobile in transcription of intracisternal A particle (IAP) elements, the population, were unmethylated (Simmen et al. 1999). which constitute a homogeneous and transpositionally This is the opposite of expectation, but may represent a active family of elements. In embryos lacking DNMT1, frequent situation in invertebrates, which account for transcription of IAP elements is massively induced, ar- >95% of animal species (Tweedie et al. 1997). guing that methylation is normally responsible for their Colonization of the genome by transposable elements repression (Walsh et al. 1998). Derepression of LINE can only occur in the germ-cell lineage because somatic (Woodcock et al. 1997) and SINE (Liu et al. 1994) pro- transposition events leave no heritable trace. Paradoxi- moters in the human genome also occurs when DNA cally, transposable elements are often transcriptionally methylation is reduced. The most abundant SINE in the active and unmethylated in germ cells and totipotent ES human genome is the Alu family, which consists of sev- cells (for review, see Bird 1997). IAP elements, for ex- eral hundred thousand elements (Smit 1999). Only a tiny ample, become unmethylated during the gonial prolif- minority of elements are capable of transposition (<1%), eration phase, when primordial germ cell number in- but many carry functional promoters. Interestingly, creases from ∼75 to ∼25,000 (Walsh et al. 1998). The fre- these promoters can be activated by stress of various quent absence of DNA methylation in germ cells, when kinds without altering DNA methylation (Liu et al. transposition can do long-term damage (Malik et al. 1995; Chu et al. 1998), although artificial demethylation 1999), contrasts with its repressive presence in somatic also stimulates expression. cells, where transposition would be an evolutionary dead The biological significance of transposable-element re- end. It is too early to discount the possibility that trans- pression is uncertain. Two kinds of explanation have poson promoters, most of which belong to degenerate been discussed: either that repression is required to pre- elements that are incapable of transposition, must be vent DNA damage due to unconstrained transposition silenced to suppress transcriptional noise. (the genome defence model; Yoder et al. 1997); or that transcription of a large excess of irrelevant promoters Mechanisms of DNA methylation-mediated would constitute an unacceptable level of transcrip- transcriptional repression tional noise that would interfere with gene expression programs (Bird 1995). Increased transcription of elements Why does DNA methylation interfere with transcrip- in human and mouse cells has not so far been found to tion? Two modes of repression can be envisaged, and it is lead to increased transposition. In undermethylated can- likely that both are biologically relevant. The first mode cer cells that show transposon promoter activity, for ex- involves direct interference of the methyl group in bind- ample, mutations caused by transposition are exceed- ing of a protein to its cognate DNA sequence (Fig. 1). ingly rare. It has, however, been claimed that rampant Many factors are known to bind CpG-containing se- transposition and reduced methylation are linked in the quences, and some of these fail to bind when the CpG is case of an interspecific hybrid marsupial (Waugh O’Neill methylated. Strong evidence for involvement of this et al. 1998). The hybrid wallaby concerned was found to mechanism in gene regulation comes from studies of the contain an abundant transposable element near the cen- role of the CTCF protein in imprinting at the H19/Igf2 tromeres of one parental chromosome set, but not the locus in mice (Bell and Felsenfeld 2000; Hark et al. 2000; other. Surprisingly, this element could not be detected in Szabo et al. 2000; Holmgren et al. 2001). CTCF is asso- either of the presumed parent species, and was therefore ciated with transcriptional domain boundaries (Bell et al. hypothesized to have been assembled from related frag- 1999) and can insulate a promoter from the influence of ments in the parental genomes following fertilization. It remote enhancers. The maternally derived copy of the was suggested that, because of perceived depression of Igf2 gene is silent owing to the binding of CTCF between

GENES & DEVELOPMENT 11 Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

Bird

Figure 1. Mechanisms of transcriptional repression by DNA methylation. A stretch of nucleosomal DNA is shown with all CpGs methylated (red circles). Below the diagram is a transcription factor that is un- able to bind its recognition site when a methylated CpG is within it. Many tran- scription factors are repelled by methyl- ation, including the boundary element protein CTCF (see text). Above the line are protein complexes that can be attracted by methylation, including the methyl-CpG- binding protein MeCP2 (plus the Sin3A histone deacetylase complex), the MeCP1 complex comprising MBD2 plus the NuRD corepressor complex, and the un- characterized MBD1 and Kaiso complexes. MeCP2 and MBD1 are chromosome- bound proteins, whereas MeCP1 may be less tightly bound. Kaiso has not yet been shown to associate with methylated sites in vivo.

its promoter and a downstream enhancer. At the pater- Excluding DNA methylation by denying access nal locus, however, these CpG-rich binding sites are The preceding discussion has considered some mecha- methylated, preventing CTCF binding and thereby al- nistic aspects of de novo DNA methylation and its bio- lowing the downstream enhancer to activate Igf2 expres- logical consequences. Although methylation affects sion. Although there is evidence that H19/Igf2 imprint- most of the mammalian genome, it is conspicuously ab- ing involves additional processes (Ferguson-Smith and sent from certain regions. Ways in which these non- Surani 2001), the role of CTCF represents one of the methylated domains may arise will now be considered. clearest examples of transcriptional regulation by DNA A simple mechanism for creating a nonmethylated do- methylation. main within an otherwise densely methylated genome is The second mode of repression is opposite to the first, to mask a stretch of DNA by protein binding. The DNA- as it involves proteins that are attracted to, rather than binding protein would accomplish this passive demeth- repelled by, methyl-CpG (Fig. 1). A family of five methyl- ylation by, for example, sterically excluding DNMTs CpG-binding proteins has been characterized that each (Bird 1986). The feasibility of this mechanism has been contains a region closely related to the methyl-CpG- verified using an artificially methylated episome con- binding domain (MBD) of MeCP2 (Nan et al. 1993, 1997; taining EBNA1 or lac repressor binding sites (Hsieh Cross et al. 1997; Hendrich and Bird 1998). Four of these 1999a; Lin et al. 2000). The idea that CpG islands are proteins—MBD1, MBD2, MBD3, and MeCP2—have entirely attributable to exclusion of this kind is in doubt, been implicated in methylation-dependent repression of however, as in vivo footprinting and nuclease accessibil- transcription (for review, see Bird and Wolffe 1999). An ity studies show CpG islands to be more accessible to unrelated protein Kaiso has also recently been shown to proteins (nucleases) than bulk genomic DNA, not less bind methylated DNA and bring about methylation-de- (Tazi and Bird 1990). Of course, it is possible that pro- pendent repression in model systems (Prokhortchouk et tection is only present at the transient embryonic stage al. 2001). In vitro, Kaiso requires a 5Ј m5CGm5CG motif, when mammalian de novo methylation occurs and has and binding is highly dependent on the presence of meth- therefore escaped detection. A protein that is reported to ylation. The presence of multiple methyl-CpG-binding bind unmethylated CpGs might be a candidate CpG-is- proteins with repressive properties supports the argu- land protector (Voo et al. 2000). ment that these may be important mediators of the methylation signal, but their involvement in specific processes that require transduction of the DNA methyl- Immunity to DNA methylation caused ation signal has yet to be shown. Targeted mutation of by transcriptionally active chromatin: the gene for MeCP2 is, however, associated with neuro- the origin of unmethylated CpG islands logical dysfunction in humans and mice (Amir et al. 1999; Chen et al. 2001; Guy et al. 2001), and mutation of Many of the known biological effects of DNA methyl- the mouse Mbd2 gene leads to a maternal behavior de- ation are associated with CpG islands. It has been argued fect (Hendrich et al. 2001). above that their methylation in the early embryo follows

12 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

DNA methylation and epigenetic memory silencing events that are likely to be DNA methylation- Why should active promoter regions escape de novo independent. If transcriptional silence indeed triggers methylation? CpG islands often colocalize with origins DNA methylation, then the corollary is that promoter of DNA replication (Delgado et al. 1998), and, according activity early in development should create a methyl- to one speculation, an early replication intermediate cre- ation-free CpG island (Fig. 2). In other words, unmethyl- ates the DNA methylation-free footprint (Antequera and ated CpG islands might be footprints of embryonic pro- Bird 1999). A more direct (but not mutually exclusive) moter activity. An obvious prediction of this model is mechanism would involve the sensing of chromatin that all unmethylated CpG islands, including those at states by the de novo methylation system as discussed promoters of highly tissue-specifically expressed genes, above. Whereas histone H3 tails modified by methyl- should contain promoters that function during early de- ation on Lys 9 might recruit DNA methyltransferases velopment when the methylation memory system is (Tamaru and Selker 2001), modifications associated with most active. Although very limited, the data so far favor active chromatin, such as acetylation of H3 or H4 or this theory, because a CpG-island promoter whose prod- methylation of Lys 4 of histone H3, may actively exclude uct RNA is not expected to occur in the early embryo these enzymes. Biochemical evidence addressing this is- (␣-globin) is nevertheless expressed, whereas transcripts sue is eagerly awaited. from a CpG-deficient promoter (␤-globin) are not de- tected (Daniels et al. 1997). Similarly, expression of the Active demethylation of DNA 68k neurofilament gene, which has a CpG-island pro- moter, was detected in ES cells, but opsin and casein Protection against de novo methylation by bound pro- genes, which are CpG-deficient genes, appeared to be teins or chromatin can ensure that DNA methylation silent (MacLeod et al. 1998). never reaches a DNA sequence domain. Unmethylated

Figure 2. A hypothetical scenario relating embryonic transcriptional activity to DNA methylation status in mammals. Starting from a notional transcription ground state, embryonic demethylation leads to substitution of methylated sites (red circles) by nonmethyl- ated sites (yellow circles). Two alternative fates are then envisaged: either transcription persists leading to restoration of the unmeth- ylated CpG island (bracket) flanked by methylated non-island-flanking DNA (pink arrows); or transcription is extinguished by other mechanisms in the embryo and this invites de novo methylation of the CpG island and its flanks. In this way the activity of embryonic promoters is imprinted for the duration of that somatic lifetime.

GENES & DEVELOPMENT 13 Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

Bird domains could also arise by actively removing the modi- choose such different routes to the same end? An intrigu- fication from DNA. This so-called active demethylation ing possibility is that the parental struggle over maternal could be accomplished either by the thermodynamically resources for the embryo that is thought to underlie ge- unfavorable breakage of the carbon—carbon bond that nomic imprinting (Moore and Haig 1991) may be in- links the pyrimidine to its methyl group, or by a repair- volved. The oocyte may be equipped to directly disarm like process that excises the m5C base or nucleoside, the sperm genome of methylation imprints that might leading to its replacement with C (Kress et al. 2001). overexploit maternal resources (Reik and Walter 2001). It Several laboratories have striven to isolate demethylase is even possible that the paternal genome, in delayed enzymes (for review, see Wolffe et al. 1999). The most retaliation, may organize a campaign of interference impressive catalytic activity was shown by a fraction with the maintenance methylation (e.g., by exporting derived from human cells (Ramchandani et al. 1999) that maternal DNMTs to the cytoplasm). The extraordinary was subsequently identified as MBD2 (Bhattacharya et need for an oocyte variant of DNMT1 to translocate into al. 1999). The expressed protein reportedly showed ro- the nucleus for only one cleavage cycle (the doubling bust demethylation in vitro in the absence of added co- from 8 to 16 cells; Howell et al. 2001) could represent factors and released methanol as a by-product. Attempts maternal measures to compensate for interference of this to observe this property of MBD2 in other laboratories kind. have not been successful. A cell extract showing demethylase activity was de- Consequences of methylation loss: gene activation tected in rat myoblast cells (Weiss et al. 1996). Initial during development indications that the reaction was RNA-dependent were not sustained upon further enrichment of the activity Interest in DNA methylation has long been fueled by the (Swisher et al. 1998). An RNA-containing demethylating notion that strategic loss of methyl groups during devel- complex was, however, reported in chicken cells (Jost et opment could lead to activation of specific genes in the al. 1997, 1999). These investigators searched for proteins appropriate lineage. As has been emphasized (Walsh and with m5C-DNA glycosylase activity and identified the Bestor 1999), much of the evidence for this scenario is previously known thymine DNA glycosylase TDG, inconclusive, but recent studies have revived the idea. In which can remove the pyrimidine base from T:G or U:G the frog, gene expression is suppressed from fertilization mismatches (Zhu et al. 2000b). MBD4, an unrelated until the mid-blastula stage (∼5000 cells), at which time DNA glycosylase with similar properties, was also found transcription is activated. Inhibition of DNMT1 using an to be active against m5C:G pairs (Zhu et al. 2000a). As antisense strategy caused reduced methylation and pre- the efficiency of these reactions was much lower than mature activation of certain genes, suggesting a direct that seen with the cognate mismatched substrates, it role for DNA methylation in maintaining their early si- might be argued that the m5C glycosylase activity rep- lence prior to the blastula stage (Stancheva and Meehan resents a minor side reaction of little in vivo signifi- 2000). Deletion of the Dnmt1 gene in cultured somatic cance. Set against this is evidence that stable expression cells of the mouse also caused widespread gene activa- of a chicken TDG results in significant activation and tion (Jackson-Grusby et al. 2001). About 10% of all genes concomitant demethylation of a reporter gene driven by detected using microarray technology were activated, a methylated ecdysone-retinoic acid-responsive pro- whereas only 1%–2% were down-regulated. Some of the moter (Zhu et al. 2001). The normally silent reporter up-regulated genes are normally only expressed in termi- could also be activated by demethylation with 5-azacyti- nally differentiated cells. These findings raise the possi- dine, but generalized demethylation of the genome was bility that DNA methylation contributes to silencing of not observed in TDG transfected cells. Previous studies tissue-specific genes in nonexpressing cells, and they showed an association between retinoid receptors and confirm DNA methylation as a global repressor of gene TDG, and implicated TDG in transcriptional activation expression. The scenario has been modeled using an ar- (Um et al. 1998). Time will tell if the stimulation of tificial construct that contained a DNA sequence ca- retinoid-responsive promoters by TDG depends on its pable of excluding methylation locally during early de- demethylating activity. velopment (Siegfried et al. 1999). When this sequence The need to isolate demethylating enzymes has be- was present, the reporter gene stayed unmethylated dur- come more acute with the finding that the paternal ge- ing development, and widespread expression occurred. nome is subject to active demethylation soon after fer- Deletion of the element in situ in the early embryo led to tilization (Mayer et al. 2000; Oswald et al. 2000). Similar methylation of the reporter gene and concomitant si- processes have been reported in pig and bovine embryos lencing in several adult tissues. (Bourc’his et al. 2001; Kang et al. 2001a,b). This dramatic A subtle potential role for loss of methylation at a illustration of methylation loss in the absence of DNA specific gene has been reported for the rat tyrosine ami- replication raises questions about the prevalence of de- notransferase gene (Thomassin et al. 2001). When a methylation by this mechanism. Interestingly, the ma- methylated form of this gene is induced by glucocorti- ternal genome, which also demethylates during early coids, delayed demethylation occurs at specific sites in mouse development, does so by a different mechanism: an enhancer and additional DNA-associated factors are passive failure to methylate progeny stands (Rougier et subsequently recruited. Demethylation (whether active al. 1998). Why should maternal and paternal genomes or passive is not known) persists after the wave of TAT

14 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

DNA methylation and epigenetic memory expression has subsided, and reinduction of the silent posed relationship between genome integrity and DNA gene by a further hormone treatment is significantly methylation that will need to be addressed by further stronger as a result. This system provides a model for a research. DNA methylation-mediated memory of the first hor- mone induction (Kress et al. 2001). Its significance is Developmental memory: DNA methylation somewhat less certain in normal development, however, and Polycomb/trithorax complexes because demethylation of these sites occurs before the as interchangeable systems gene becomes hormone-inducible. There is suggestive evidence that programmed rear- The foregoing discussion has highlighted features of the rangement of the immunoglobulin genes during B-cell DNA methylation system in mammals that resemble development may involve DNA methylation (Mosto- another established system of cellular memory: Pc-G/ slavsky et al. 1998). Demethylation of one of the two trx. The final section of the review will compare the two parentally derived alleles of the kappa light chain gene is systems. The credentials of Pc-G/trx protein complexes observed in small pre-B cells, and there is evidence that as an epigenetic system in development are compelling this early loss of methylation predisposes the affected (Paro et al. 1998; Pirrotta 1999; Francis and Kingston allele to rearrangement. By precluding rearrangement of 2001). This multiprotein assembly is targeted to specific one allele, differential DNA methylation may help to regions of the genome where it effectively freezes the explain allelic exclusion at the kappa chain locus. It is embryonic expression status of a gene, be it active or not certain whether transcriptional regulation per se inactive, and propagates that state stably through devel- plays a role, although the process is dependent on the opment. Elegant experiments with model gene con- intronic and 3Ј kappa gene enhancers. structs have shown that brief activation (or inactivation) of a promoter during early Drosophila development leads to stable activity (or inactivity) thereafter (Cavalli and Loss of genome integrity as a consequence of DNA Paro 1998, 1999; Poux et al. 2001). Attempts to alter methylation loss? expression at most other stages of development were un- Early studies with the DNA-methylation inhibitor successful, indicating that there is a window of time dur- 5-azacytidine revealed bizarre chromosomal rearrange- ing which transcription patterns can be committed to ments in treated cultured cells (Viegas-Pequignot and developmental memory. The Pc-G/trx system is reactive Dutrillaux 1976). Although these findings might be at- rather than proactive, as the setting up of segment-spe- tributed to the effects of reduced DNA methylation, they cific patterns of active genes is not disrupted by muta- could also be a result of the chemical reactivity of the tions in Pc-G group genes. Only the capacity to sustain incorporated base analog, in particular, its ability to the patterns is lost in the mutants. This ability to copy cross-link proteins to DNA (Juttermann et al. 1994). The and propagate the expression patterns without influenc- former possibility is supported somewhat by the finding ing or perturbing them makes this a subtle and flexible that mitogen-stimulated lymphocytes from patients memory system. Little is known, however, about the with mutations in DNMT3B show very similar chromo- mechanisms responsible for the heritability of Pc-G/trx. some rearrangements, involving coalescence of centro- What do Pc-G/trx and DNA methylation in mammals meric regions that contain methylation-deficient repeti- have in common? First, both systems are able to repress tive sequences (Jeanpierre et al. 1993; Xu et al. 1999). transcription in a heritable manner. Second, both appear Oddly, the rearrangements are not seen in cells of the to be reactive in that they lock in expression states that patients, despite similar hypomethylation of these re- they played no part in setting up (e.g., DNA methylation gions. It seems that loss of genomic integrity is not an in viral genome silencing and CpG-island methylation obligatory consequence of hypomethylation of juxtacen- on the X chromosome). Third, both are activated prima- tromeric repeat elements. rily during a discrete window of time in early develop- At a finer level, two laboratories have examined the ment. Thus, like Pc-G/trx, DNA methylation has the effects of greatly reduced DNA methylation levels on properties of a developmental memory. mutation rates in mouse embryonic stem cells, with What is memorized by DNA methylation? Arguably, somewhat differing results. In one study, the mutation its major role is to stably demarkate by its absence a set rate at two endogenous loci was found to have increased of embryonically active promoters, namely, CpG is- ∼10-fold compared to the same loci in wild-type cells lands, so that they remain potentially active throughout (Chen et al. 1998), suggesting that lack of methylation development and adulthood. At the same time, regions predisposed to aberrant recombination events. A second devoid of promoter activity in the embryo become meth- study examined transgenes of exogenous origin using a ylated and carry this repressive influence with them selection system to detect mutations (Chan et al. 2001). through development. The degree of repression may be This allowed screening of large numbers of mutations at weak or strong depending on the density of methylation two independent loci, but neither point mutations nor (Boyes and Bird 1992; Hsieh 1994). Thus, CpG islands genomic rearrangements were increased under condi- that are silenced by other mechanisms during embryo- tions of limiting DNA methylation. In fact, mutations genesis would acquire dense methylation leading to ir- appeared to be suppressed by genomic hypomethylation. reversible silencing. When, however, the density of These inconsistencies raise questions about the pro- methylated CpGs is low, as it is in most of the genome,

GENES & DEVELOPMENT 15 Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

Bird repression is likely to be weak and may be overcome by intervene to silence genes that are actively transcribed, the presence of strong activators. Weak repression of tis- but only affects genes that have already been shut down sue-specific genes (e.g., ␤-globin) that are embedded in by other means. There is reason to believe that transcrip- regions of low-density methylation may contribute to tional activity may somehow imprint the methylation- their silence in inappropriate tissues. free status of CpG islands. The involvement of DNA It is proposed here that DNA methylation and Pc-G/ methylation in inactivation of transposable elements trx are alternative systems of cellular memory that are could likewise be due to its capacity for stabilizing the interchangeable over evolutionary time. In C. elegans transcriptional shutdown organized by other systems. and Drosophila, for example, Pc-G group proteins Parallels between these emerging attributes of DNA (Birchler et al. 2000) have been implicated in silencing methylation and the Pc-G system in Drosophila suggest of repetitive-element transcription in somatic cells, that both are mechanisms for sensing and propagating whereas DNA methylation may play this role in mam- cellular memory. mals (Yoder et al. 1997). The involvement of DNA meth- ylation in genome defence may, therefore, be to memo- rize the silent state of elements imposed by primary ge- Acknowledgments nome defence systems. The most striking evidence for I am grateful to Eric Selker, Bernard Ramsahoye, Helle Jør- interchangeability is the finding that X chromosome in- gensen, Brian Hendrich, and Catherine Millar for comments on activation in extraembryonic tissues of the mouse de- the manuscript. Research by A.B. is supported by The Wellcome pends on the polycomb group protein Eed. Loss of the Trust. eed gene leads to reactivation of the inactive X in extra- embryonic tissue, but has no effect in somatic cell types (Wang et al. 2001). In contrast, Dnmt1 mutations reac- References tivate the inactive X of the embryo proper, but not the extraembryonic inactive X (Sado et al. 2000). The finding Amir, R.E., Van den Veyver, I.B., Wan, M., Tran, C.Q., Francke, U., and Zoghbi, H.Y. 1999. Rett syndrome is caused by mu- that certain CpG islands on the inactive X chromosome tations in X-linked MECP2, encoding methyl-CpG-binding are methylated in somatic cells but not in extraembry- protein 2. Nat. Genet. 23: 185–188. onic tissues (Iida et al. 1994) fits with the view that Antequera, F. and Bird, A. 1993. Number of CpG islands and methylation replaces Pc-G in somatic tissues. Therefore, genes in human and mouse. Proc. Natl. Acad. Sci. even within a single species, it appears that different 90: 11995–11999. tissues employ Pc-G/trx and DNA methylation inter- ———. 1999. CpG islands as genomic footprints of promoters changeably. From an evolutionary perspective, it is pos- that are associated with replication origins. Curr. Biol. sible that varying degrees of functional substitution by 9: R661–R667. Pc-G (or vice versa) can explain the dramatic differences Antequera, F., Boyes, J., and Bird, A. 1990. High levels of de novo between DNA methylation levels across animal species. methylation and altered chromatin structure at CpG islands in cell lines. Cell 62: 503–514. Bachman, K.E., Rountree, M.R., and Baylin, S.B. 2001. Dnmt3a Concluding remarks and Dnmt3b are transcriptional repressors that exhibit unique localization properties to heterochromatin. J. Biol. Our understanding of the relationship between DNA Chem. 276: 32282–32287. methylation and transcriptional control is growing fast, Bannister, A.J., Zegerman, P., Partridge, J.F., Miska, E.A., Thom- but is still far from complete. Ongoing biochemical as, J.O., Allshire, R.C., and Kouzarides, T. 2001. Selective analysis of the growing number of components of the recognition of methylated lysine 9 on histone H3 by the HP1 DNA methylation system (and their partners), coupled chromo domain. Nature 410: 120–124. with genetic approaches, will strengthen the links be- Baylin, S.B. and Herman, J.G. 2000. DNA hypermethylation in tween DNA methylation and mainstream transcrip- tumorigenesis: Epigenetics joins genetics. Trends Genet. 16: 168–174. tional mechanisms. Regulation of gene expression is Beard, C., Li, E., and Jaenisch, R. 1995. Loss of methylation complex (Lemon and Tjian 2000), and the emerging evi- activates Xist in somatic but not in embryonic cells. Genes dence hints that the roles of DNA methylation will be & Dev. 9: 2325–2334. too. It may be unrealistic to expect that any unified Bell, A.C. and Felsenfeld, G. 2000. Methylation of a CTCF-de- theory will encompass all the biological consequences of pendent boundary controls imprinted expression of the Igf2 DNA methylation. gene. Nature 405: 482–485. Least understood are the mechanisms by which meth- Bell, A.C., West, A.G., and Felsenfeld, G. 1999. The protein ylation patterns are generated. Following consideration CTCF is required for the enhancer blocking activity of ver- of the criteria for attracting and repelling DNA methyl- tebrate insulators. Cell 98: 387–396. ation, this review has entertained the possibility that a Bender, J. 2001. A vicious cycle. RNA silencing and DNA meth- ylation in plants. Cell 106: 129–132. primary function of de novo DNA methylation is to Bestor, T.H. 1992. Activation of mammalian DNA methyl- memorize patterns of embryonic gene activity, creating transferase by cleavage of a Zn binding regulatory domain. CpG islands that are competent for transcription EMBO J. 11: 2611–2617. throughout development, or their antithesis, regions ———. 2000. The DNA methyltransferases of mammals. Hum. that are methylated and transcriptionally incompetent. Mol. Genet. 9: 2395–2402. The idea depends on evidence that methylation does not Bestor, T. and Tycko, B. 1996. Creation of genomic methylation

16 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

DNA methylation and epigenetic memory

patterns. Nat. Genet. 12: 363–367. Colot, V., Maloisel, I.L., and Rossignol, J.-L. 1996. Interchromo- Bhattacharya, S.K., Ramchandani, S., Cervoni, N., and Szyf, M. somal transfer of epigenetic states in Ascobolus: Transfer of 1999. A mammalian protein with specific demethylase ac- DNA methylation is mechanistically related to homologous tivity for mCpG DNA. Nature 397: 579–583. recombination. Cell 86: 855–864. Birchler, J.A., Bhadra, M.P., and Bhadra, U. 2000. Making noise Cosma, M.P., Tanaka, T., and Nasmyth, K. 1999. Ordered re- about silence: Repression of repeated genes in animals. Curr. cruitment of transcription and chromatin remodeling factors Opin. Genet. Dev. 10: 211–216. to a cell cycle- and developmentally-regulated promoter. Bird, A.P. 1986. CpG-rich islands and the function of DNA Cell 97: 299–311. methylation. Nature 321: 209–213. Costello, J.F. and Plass, C. 2001. Methylation matters. J. Med. ———. 1987. CpG islands as gene markers in the vertebrate Genet. 38: 285–303. nucleus. Trends Genet. 3: 342–347. Cross, S.H., Meehan, R.R., Nan, X., and Bird, A. 1997. A com- ———. 1995. Gene number, noise reduction and biological ponent of the transcriptional repressor MeCP1 is related to complexity. Trends Genet. 11: 94–100. mammalian DNA methyltransferase and trithorax-like pro- ———. 1997. Does DNA methylation control transposition of tein. Nat. Genet. 16: 256–259. selfish elements in the germline. Trends Genet. 13: 469–470. Cuadrado, M., Sacristan, M., and Antequera, F. 2001. Species- Bird, A. and Wolffe, A.P. 1999. Methylation-induced repres- specific organization of CpG island promoters at mamma- sion—Belts, braces and chromatin. Cell 99: 451–454. lian homologous genes. EMBO Rep. 2: 586–592. Bird, A.P., Taggart, M.H., and Smith, B.A. 1979. Methylated and Daniels, R., Lowell, S., Bolton, V., and Monk, M. 1997. Tran- unmethylated DNA compartments in the sea urchin ge- scription of tissue-specific genes in human preimplantation nome. Cell 17: 889–901. embryos. Hum. Reprod. 12: 2251–2256. Bird, A.P., Taggart, M.H., Nicholls, R.D., and Higgs, D.R. 1987. Delgado, S., Gomez, M., Bird, A., and Antequera, F. 1998. Ini- Non-methylated CpG-rich islands at the human ␣-globin lo- tiation of DNA replication at CpG islands in mammalian cus: Implications for evolution of the ␣-globin pseudogene. chromosomes. EMBO J. 17: 2426–2435. EMBO J. 6: 999–1004. Dennis, K., Fan, T., Geiman, T., Yan, Q., and Muegge, K. 2001. Birger, Y., Shemer, R., Perk, J., and Razin, A. 1999. The imprint- Lsh, a member of the SNF2 family, is required for genome- ing box of the mouse Igf2r gene. Nature 397: 84–88. wide methylation. Genes & Dev. 15: 2940–2944. Bourc’his, D., Le Bourhis, D., Patin, D., Niveleau, A., Comiz- De Smet, C., De Backer, O., Faraoni, I., Lurquin, C., Brasseur, F., zoli, P., Renard, J., and Viegas-Pequignot, E. 2001. Delayed and Boon, T. 1996. The activation of human gene MAGE-1 and incomplete reprogramming of chromosome methylation in tumor cells is correlated with genome-wide demethyl- patterns in bovine cloned embryos. Curr. Biol. ation. Proc. Natl. Acad. Sci. 93: 7149–7153. 11: 1542–1546. De Smet, C., Lurquin, C., Lethe, B., Martelange, V., and Boon, T. Boyes, J. and Bird, A. 1992. Repression of genes by DNA meth- 1999. DNA methylation is the primary silencing mechanism ylation depends on CpG density and promoter strength: Evi- for a set of germ line- and tumor-specific genes with a CpG- dence for involvement of a methyl-CpG binding protein. rich promoter. Mol. Cell. Biol. 11: 7327–7335. EMBO J. 11: 327–333. Ehrlich, M. 1982. Amount and distribution of 5-methycytosine Brandeis, M., Frank, D., Keshet, I., Siegried, Z., Mendelsohn, M., in human DNA from different types of tissues or cells. Nemes, A., Temper, V., Razin, A., and Cedar, H. 1994. Sp1 Nucleic Acids Res. 10: 2709–2721. elements protect a CpG island from de novo methylation. Ferguson-Smith, A.C. and Surani, M.A. 2001. Imprinting and Nature 371: 435–438. the epigenetic asymmetry between parental genomes. Sci- Cavalli, G. and Paro, R. 1998. The Drosophila Fab-7 chromo- ence 293: 1086–1089. somal element conveys epigenetic inheritance during mito- Finnegan, E.J., Peacock, W.J., and Dennis, E.S. 2000. DNA meth- sis and meiosis. Cell 93: 505–518. ylation, a key regulator of plant development and other pro- ———. 1999. Epigenetic inheritance of active chromatin after cesses. Curr. Opin. Genet. Dev. 10: 217–223. removal of the main transactivator. Science 286: 955–958. Flatau, E., Gonzales, F.A., Michalowsky, L.A., and Jones, P.A. Chan, M.F., van Amerongen, R., Nijjar, T., Cuppen, E., Jones, 1984. DNA methylation in 5-Aza-2Ј-deoxycytidine-resistant P.A., and Laird, P.W. 2001. Reduced rates of gene loss, gene variants of C3H 10T1/2 C18 cells. Mol. Cell. Biol. 4: 2098– silencing, and gene mutation in dnmt1-deficient embryonic 2102. stem cells. Mol. Cell. Biol. 21: 7587–7600. Francis, N.J. and Kingston, R.E. 2001. Mechanisms of transcrip- Chen, R.Z., Pettersson, U., Beard, C., Jackson-Grusby, L., and tional memory. Nat. Rev. Mol. Cell. Biol. 2: 409–421. Jaenisch, R. 1998. DNA hypomethylation leads to elevated Fuks, F., Burgers, W.A., Brehm, A., Hughes-Davies, L., and mutation rates. Nature 395: 89–93. Kouzarides, T. 2000. DNA methyltransferase Dnmt1 associ- Chen, R.Z., Akbarian, S., Tudor, M., and Jaenisch, R. 2001. De- ates with histone deacetylase activity. Nat. Genet. 24: 88– ficiency of methyl-CpG binding protein-2 in CNS neurons 91. results in a Rett-like phenotype in mice. Nat. Genet. Fuks, F., Burgers, W.A., Godin, N., Kasai, M., and Kouzarides, T. 27: 327–331. 2001. Dnmt3a binds deacetylases and is recruited by a se- Chu, W.M., Ballard, R., Carpick, B.W., Williams, B.R., and quence-specific repressor to silence transcription. EMBO J. Schmid, C.W. 1998. Potential Alu function: Regulation of 20: 2536–2544. the activity of double-stranded RNA-activated kinase PKR. Garrick, D., Fiering, S., Martin, D.I., and Whitelaw, E. 1998. Mol. Cell. Biol. 18: 58–68. Repeat-induced gene silencing in mammals. Nat. Genet. Cogoni, C., Irelan, J.T., Schumacher, M., Schmidhauser, T.J., 18: 56–59. Selker, E.U., and Macino, G. 1996. Transgene silencing of Gautsch, J.W. and Wilson, M.C. 1983. Delayed de novo meth- the al-1 gene in vegetative cells of Neurospora is mediated by ylation in teratocarcinoma cells suggests additional tissue- a cytoplasmic effector and does not depend on DNA–DNA specific mechanisms for controlling gene expression. Nature interactions or DNA methylation. EMBO J. 15: 3153–3163. 301: 32–37. Colot, V. and Rossignol, J.L. 1999. Eukaryotic DNA methyl- Gibbons, R.J., McDowell, T.L., Raman, S., O’Rourke, D.M., ation as an evolutionary device. BioEssays 21: 402–411. Garrick, D., Ayyub, H., and Higgs, D.R. 2000. Mutations in

GENES & DEVELOPMENT 17 Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

Bird

ATRX, encoding a SWI/SNF-like protein, cause diverse agome, Y. 1994. The CpG island of the FMR-1 gene is meth- changes in the pattern of DNA methylation. Nat. Genet. ylated differently among embryonic tissues: Implication for 24: 368–371. prenatal diagnosis. Hum. Reprod. 9: 1471–1473. Gowher, H. and Jeltsch, A. 2001. Enzymatic properties of re- Issa, J.P. 2000. CpG-island methylation in aging and cancer. combinant Dnmt3a DNA methyltransferase from mouse: Curr. Top. Microbiol. Immunol. 249: 101–118. The enzyme modifies DNA in a non-processive manner and Jackson-Grusby, L., Beard, C., Possemat, R., Tudor, M., Fam- also methylates non-CpG [correction of non-CpA] sites. J. brough, D., Csankovszki, G., Dausman, J., Lee, P., Wilson, Mol. Biol. 309: 1201–1208. C., Lander, E., et al. 2001. Loss of genomic methylation Gowher, H., Leismann, O., and Jeltsch, A. 2000. DNA of Dro- causes p53-dependent apoptosis and epigenetic deregulation. sophila melanogaster contains 5-methylcytosine. EMBO J. Nat. Genet. 27: 31–39. 19: 6918–6923. Jaenisch, R. 1997. DNA methylation and imprinting: Why Graves, J.A. 1982. 5-Azacytidine-induced re-expression of alle- bother? Trends Genet. 13: 323–329. les on the inactive X chromosome in a hybrid mouse cell Jaenisch, R., Harbers, K., Jahner, D., Stewart, C., and Stuhl- line. Exp. Cell Res. 141: 99–105. mann, H. 1982. DNA methylation, retroviruses, and em- Gurdon, J.B. 1999. Genetic reprogramming following nuclear bryogenesis. J. Cell Biochem. 20: 331–336. transplantation in Amphibia. Semin. Cell Dev. Biol. Jaenisch, R., Schnieke, A., and Harbers, K. 1985. Treatment of 10: 239–243. mice with 5-azacytidine efficiently activates silent retroviral Guy, J., Hendrich, B., Holmes, M., Martin, J.E., and Bird, A. genomes in different tissues. Proc. Natl. Acad. Sci. 82: 1451– 2001. A mouse Mecp2-null mutation causes neurological 1455. symptoms that mimic Rett syndrome. Nat. Genet. (in press). Jeanpierre, M., Turleau, C., Aurias, A., Prieur, M., Ledeist, F., Hansen, R.S., Stoger, R., Wijmenga, C., Stanek, A.M., Canfield, Fischer, A., and Viegas-Pequignot, E. 1993. An embryonic- T.K., Luo, P., Matarazzo, M.R., D’Esposito, M., Feil, R., like methylation pattern of classical satellite DNA is ob- Gimelli, G., et al. 2000. Escape from gene silencing in ICF served in ICF syndrome. Hum. Mol. Genet. 2: 731–735. syndrome: Evidence for advanced replication time as a major Jeddeloh, J.A., Stokes, T.L., and Richards, E.J. 1999. Mainte- determinant. Hum. Mol. Genet. 9: 2575–2587. nance of genomic methylation requires a SW12/SNF2-like Hark, A.T., Schoenherr, C.J., Katz, D.J., Ingram, R.S., Levorse, protein. Nat. Genet. 22: 94–97. J.M., and Tilghman, S.M. 2000. CTCF mediates methyl- Jenuwein, T. and Allis, C.D. 2001. Translating the histone code. ation-sensitive enhancer-blocking activity at the H19/Igf2 Science 293: 1074–1080. locus. Nature 405: 486–489. Jones, P.A. and Takai, D. 2001. The role of DNA methylation in Harris, M. 1982. Induction of thymidine kinase in enzyme-de- mammalian epigenetics. Science 293: 1068–1070. ficient Chinese hamster cells. Cell 29: 483–492. Jones, P.A. and Taylor, S.M. 1980. Cellular differentiation, cy- Hendrich, B. and Bird, A. 1998. Identification and characteriza- tidine analogues and DNA methylation. Cell 20: 85–93. tion of a family of mammalian methyl-CpG binding pro- Jones, P.A., Wolkowicz, M.J., Rideout, W.M.I., Gonzales, F.A., teins. Mol. Cell. Biol. 18: 6538–6547. Marziasz, C.M., Coetzee, G.A., and Tapscott, S.J. 1990. De Hendrich, B., Guy, J., Ramsahoye, B., Wilson, V.A., and Bird, A. novo methylation of the MyoD1 CpG island during the es- 2001. Closely related proteins Mbd2 and Mbd3 play distinc- tablishment of immortal cell lines. Proc. Natl. Acad. Sci. tive but interacting roles in mouse development. Genes & 87: 6117–6121. Dev. 15: 1613–1618. Jost, J.P., Fremont, M., Siegmann, M., and Hofsteenge, J. 1997. Hertz, J.M., Schell, G., and Doerfler, W. 1999. Factors affecting The RNA moiety of chick embryo 5-methylcytosine-DNA de novo methylation of foreign DNA in mouse embryonic glycosylase targets DNA demethylation. Nucleic Acids Res. stem cells. J. Biol. Chem. 274: 24232–24240. 25: 4545–4550. Holliday, R. and Pugh, J.E. 1975. DNA modification mecha- Jost, J.P., Siegmann, M., Thiry, S., Jost, Y.C., Benjamin, D., and nisms and gene activity during development. Science Schwarz, S. 1999. A re-investigation of the ribonuclease sen- 186: 226–232. sitivity of a DNA demethylation reaction in chicken embryo Holmgren, C., Kanduri, C., Dell, G., Ward, A., Mukhopadhya, and G8 mouse myoblasts. FEBS Lett. 449: 251–254. R., Kanduri, M., Lobanenkov, V., and Ohlsson, R. 2001. CpG Juttermann, R., Li, E., and Jaenisch, R. 1994. Toxicity of 5-aza- methylation regulates the Igf2/H19 insulator. Curr. Biol. 2Ј-deoxycytidine to mammalian cells is mediated primarily 11: 1128–1130. by covalent trapping of DNA methyltransferase rather than Howell, C.Y., Bestor, T.H., Ding, F., Latham, K.E., Mertineit, C., DNA demethylation. Proc. Natl. Acad. Sci. Trasler, J.M., and Chaillet, J.R. 2001. Genomic imprinting 91: 11797–11801. disrupted by a maternal effect mutation in the Dnmt1 gene. Kafri, T., Ariel, M., Brandeis, M., Shemer, R., Urven, L., McCar- Cell 104: 829–838. rey, J., Cedar, H., and Razin, A. 1992. Developmental pattern Hsieh, C.-L. 1994. Dependence of transcriptional repression on of gene-specific DNA methylation in the mouse embryo and CpG methylation density. Mol. Cell. Biol. 14: 5487–5494. germ line. Genes & Dev. 6: 705–714. ———. 1999a. Evidence that protein binding specifies sites of Kang, Y.K., Koo, D.B., Park, J.S., Choi, Y.H., Chung, A.S., Lee, DNA demethylation. Mol. Cell. Biol. 19: 46–56. K.K., and Han, Y.M. 2001a. Aberrant methylation of donor ———. 1999b. In vivo activity of murine de novo methyltrans- genome in cloned bovine embryos. Nat. Genet. 28: 173–177. ferases, Dnmt3a and Dnmt3b. Mol. Cell. Biol. Kang, Y.K., Koo, D.B., Park, J.S., Choi, Y.H., Kim, H.N., Chang, 19: 8211–8218. W.K., Lee, K.K., and Han, Y.M. 2001b. Typical demethyl- ———. 2000. Dynamics of DNA methylation pattern. Curr. ation events in cloned pig embryos. Clues on species-specific Opin. Genet. Dev. 10: 224–228. differences in epigenetic reprogramming of cloned donor ge- Hung, M.S., Karthikeyan, N., Huang, B., Koo, H.C., Kiger, J., and nome. J. Biol. Chem. 276: 39980–39984. Shen, C.J. 1999. Drosophila proteins related to vertebrate Kass, S.U., Landsberger, N., and Wolffe, A.P. 1997a. DNA meth- DNA (5-cytosine) methyltransferases. Proc. Natl. Acad. Sci. ylation directs a time-dependent repression of transcription 96: 11940–11945. initiation. Curr. Biol. 7: 157–165. Iida, T., Nakahori, Y., Tsutsumi, O., Taketani, Y., and Nak- Kass, S.U., Pruss, D., and Wolffe, A.P. 1997b. How does DNA

18 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

DNA methylation and epigenetic memory

methylation repress transcription? Trends Genet. 13: 444– MacLeod, D., Clark, V., and Bird, A. 1999. Absence of genome- 449. wide changes in DNA methylation during development of Keohane, A.M., O’Neill, L.P., Belyaev, N.D., Lavender, J.S., and the zebrafish (Danio rerio). Nat. Genet. 23: 139–140. Turner, B.M. 1996. X-inactivation and histone H4 acetyla- Malik, H.S., Burke, W.D., and Eickbush, T.H. 1999. The age and tion in embryonic stem cells. Dev. Biol. 180: 618–630. evolution of non-LTR retrotransposable elements. Mol. Biol. Kondo, T., Bobek, M.P., Kuick, R., Lamb, B., Zhu, X., Narayan, Evol. 16: 793–805. A., Bourc’his, D., Viegas-Pequignot, E., Ehrlich, M., et al. Martienssen, R.A. and Colot, V. 2001. DNA methylation and 2000. Whole-genome methylation scan in ICF syndrome: epigenetic inheritance in plants and filamentous fungi. Sci- Hypomethylation of non-satellite DNA repeats D4Z4 and ence 293: 1070–1074. NBL2. Hum. Mol. Genet. 9: 597–604. Matzke, M., Matzke, A.J., and Kooter, J.M. 2001. RNA: Guiding Kress, C., Thomassin, H., and Grange, T. 2001. Local DNA de- gene silencing. Science 293: 1080–1083. methylation in vertebrates: How could it be performed and Mayer, W., Niveleau, A., Walter, J., Fundele, R., and Haaf, T. targeted? FEBS Lett. 494: 135–140. 2000. Demethylation of the zygotic paternal genome. Na- Laayoun, A. and Smith, S.S. 1995. Methylation of slipped du- ture 403: 501–502. plexes, snapbacks and cruciforms by human DNA(cytosine- McKeon, C., Ohkubo, H., Pastan, I., and de Crombrugghe, B. 5)methyltransferase. Nucleic Acids Res. 23: 1584–1589. 1982. Unusual methylation pattern of the ␣ 2(1) collagen Lander, E.S., Linton, L.M., Birren, B., Nusbaum, C., Zody, M.C., gene. Cell 29: 203–210. Baldwin, J., Devon, K., Dewar, K., Doyle, M., FitzHugh, W., Miao, V.P., Freitag, M., and Selker, E.U. 2000. Short TpA-rich et al. 2001. Initial sequencing and analysis of the human segments of the ␨–␩ region induce DNA methylation in Neu- genome. Nature 409: 860–921. rospora crassa. J. Mol. Biol. 300: 249–273. Lemon, B. and Tjian, R. 2000. Orchestrated response: A sym- Miniou, P., Jeanpierre, M., Blanquet, V., Sibella, V., Bonneau, phony of transcription factors for gene control. Genes & D., Herbelin, C., Fischer, A., Niveleau, A., and Viegas- Dev. 14: 2551–2569. Pequignot, E. 1994. Abnormal methylation pattern in con- Li, E., Bestor, T.H., and Jaenisch, R. 1992. Targeted mutation of stitutive and facultative (X inactive chromosome) hetero- the DNA methyltransferase gene results in embryonic le- chromatin of ICF patients. Hum. Mol. Genet. 3: 2093–2102. thality. Cell 69: 915–926. Mohandas, T., Sparkes, R.S., and Shapiro, L.J. 1981a. Reactiva- Li, E., Beard, C., and Jaenisch, R. 1993. Role for DNA methyl- tion of an inactive human X chromosome: Evidence for X ation in genomic imprinting. Nature 366: 362–365. inactivation by DNA methylation. Science 211: 393–396. Lin, I.G., Tomzynski, T.J., Ou, Q., and Hsieh, C.L. 2000. Modu- ———. 1981b. Reactivation of an inactive human X-chromo- lation of DNA binding protein affinity directly affects target some: Evidence for X-inactivation by DNA methylation. Sci- site demethylation. Mol. Cell. Biol. 20: 2343–2349. ence 211: 393–396. Liu, W.M., Maraia, R.J., Rubin, C.M., and Schmid, C.W. 1994. Monk, M., Boubelik, M., and Lehnert, S. 1987. Temporal and Alu transcrips: Cytoplasmic localisation and regulation by regional changes in DNA methylation in the embryonic, ex- DNA methylation. Nucleic Acids Res. 22: 1087–1095. traembryonic and germ cell lineages during mouse embryo Liu, W.M., Chu, W.M., Choudary, P.V., and Schmid, C.W. 1995. development. Development 99: 371–382. Cell stress and translational inhibitors transiently increase Moore, T. and Haig, D. 1991. Genomic imprinting in mamma- the abundance of mammalian SINE transcripts. Nucleic Ac- lian development: A parental tug-of-war. TIG 7: 1–4. ids Res. 23: 1758–1765. Mostoslavsky, R., Singh, N., Kirillov, A., Pelanda, R., Cedar, H., Lock, L.F., Takagi, N., and Martin, G.R. 1987. Methylation of Chess, A., and Bergman, Y. 1998. ␬ chain monoallelic de- the Hprt gene on the inactive X occurs after chromosome methylation and the establishment of allelic exclusion. inactivation. Cell 48: 39–46. Genes & Dev. 12: 1801–1811. Lorincz, M.C., Schubeler, D., Goeke, S.C., Walters, M., Grou- Mummaneni, P., Bishop, P.L., and Turker, M.S. 1993. A cis- dine, M., and Martin, D.I.K. 2000. Dynamic analysis of pro- acting element accounts for a conserved methylation pattern viral induction and de novo methylation: Implications for a upstream of the mouse adenine phosphoribosyltransferase histone deacetylase-independent, methylation density-de- gene. J. Biol. Chem. 268: 552–558. pendent mechanism of transcriptional repression. Mol. Cell. Mummaneni, P., Yates, P., Simpson, J., Rose, J., and Turker, Biol. 20: 842–850. M.S. 1998. The primary function of a redundant Sp1 binding Lyko, F., Ramsahoye, B.H., Kashevsky, H., Tudor, M., Mastran- site in the mouse aprt gene promoter is to block epigenetic gelo, M.-A., Orr-Weaver, T.L., and Jaenisch, R. 1999. Mam- gene inactivation. Nucleic Acids Res. 26: 5163–5169. malian (cytosine-5) methyltransferases cause genomic DNA Nakayama, J., Rice, J.C., Strahl, B.D., Allis, C.D., and Grewal, methylation and lethality in Drosophila. Nat. Genet. S.I. 2001. Role of histone H3 lysine 9 methylation in epige- 23: 363–366. netic control of heterochromatin assembly. Science Lyko, F., Ramsahoye, B.H., and Jaenisch, R. 2000. DNA meth- 292: 110–113. ylation in Drosophila melanogaster. Nature 408: 538–540. Nan, X., Meehan, R.R., and Bird, A. 1993. Dissection of the Lyle, R., Watanabe, D., te Vruchte, D., Lerchner, W., Smrzka, methyl-CpG binding domain from the chromosomal protein O.W., Wutz, A., Schageman, J., Hahner, L., Davies, C., and MeCP2. Nucleic Acids Res. 21: 4886–4892. Barlow, D.P. 2000. The imprinted antisense RNA at the Igf2r Nan, X., Campoy, J., and Bird, A. 1997. MeCP2 is a transcrip- locus overlaps but does not imprint Mas1. Nat. Genet. tional repressor with abundant binding sites in genomic 25: 19–21. chromatin. Cell 88: 471–481. MacLeod, D., Charlton, J., Mullins, J., and Bird, A.P. 1994. Sp1 Niwa, O., Yokota, Y., Ishida, H., and Sugahara, T. 1983. Inde- sites in the mouse aprt gene promoter are required to prevent pendent mechanisms involved in suppression of the methylation of the CpG island. Genes & Dev. 8: 2282–2292. Moloney leukaemia virus genome during differentiation of MacLeod, D., Ali, R.R., and Bird, A.P. 1998. An alternative pro- murine teratocarinoma cells. Cell 32: 1105–1113. moter in the mouse major histocompatibility cmplex class II Noma, K., Allis, C.D., and Grewal, S.I. 2001. Transitions in I-Ab gene: Implications for the origin of CpG islands. Mol. distinct histone H3 methylation patterns at the heterochro- Cell. Biol. 18: 4433–4443. matin domain boundaries. Science 293: 1150–1155.

GENES & DEVELOPMENT 19 Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

Bird

Okano, M., Xie, S., and Li, E. 1998a. Cloning and characteriza- and epigenetic reprogramming of the genome. Science tion of a family of novel mammalian DNA (cytosine-5) 293: 1093–1098. methyltransferases. Nat. Genet. 19: 219–220. Riggs, A.D. 1975. X-inactivation, differentiation and DNA ———. 1998b. Dnmt2 is not required for de novo and mainte- methylation. Cytogenet. Cell. Genet. 14: 9–25. nance methylation of viral DNA in embryonic stem cells. Riggs, A.D., Xiong, Z., Wang, L., and LeBon, J.M. 1998. Meth- Nucleic Acids Res. 26: 2536–2540. ylation dynamics, epigenetic fidelity and X chromosome Okano, M., Bell, D.W., Haber, D.A., and Li, E. 1999. DNA meth- structure. Novartis Found. Symp. 214: 214–225. yltransferases Dnmt3a and Dnmt3b are essential for de novo Robertson, K.D., Ait-Si-Ali, S., Yokochi, T., Wade, P.A., Jones, methylation and mammalian development. Cell P.L., and Wolffe, A.P. 2000. DNMT1 forms a complex with 99: 247–257. Rb, E2F1 and HDAC1 and represses transcription from E2F- Oswald, J., Engemann, S., Lane, N., Mayer, W., Olek, A., Fun- responsive promoters. Nat. Genet. 25: 338–342. dele, R., Dean, W., Reik, W., and Walter, J. 2000. Active Rougier, N., Bourc’his, D., Gomes, D.M., Niveleau, A., Plachot, demethylation of the paternal genome in the mouse zygote. M., Paldi, A., and Viegas-Pequignot, E. 1998. Chromosome Curr. Biol. 10: 475–478. methylation patterns during mammalian preimplantation Pannell, D., Osborne, C.S., Yao, S., Sukonnik, T., Pasceri, P., development. Genes & Dev. 12: 2108–2113. Karaiskakis, A., Okano, M., Li, E., Lipshitz, H.D., and Ellis, Russo, V.E.A., Martienssen, R.A., and Riggs, A.D. 1996. Epige- J. 2000. Retrovirus vector silencing is de novo methylase netic mechanisms of gene regulation. Cold Spring Harbor independent and marked by a repressive histone code. Laboratory Press, Cold Spring Harbor, NY. EMBO J. 19: 5884–5894. Sado, T., Fenner, M.H., Tan, S.S., Tam, P., Shioda, T., and Li, E. Panning, B. and Jaenisch, R. 1996. DNA hypomethylation can 2000. X inactivation in the mouse embryo deficient for activate Xist expression and silence X-linked genes. Genes & Dnmt1: Distinct effect of hypomethylation on imprinted Dev. 10: 1991–2002. and random X inactivation. Dev. Biol. 225: 294–303. Paro, R., Strutt, H., and Cavalli, G. 1998. Heritable chromatin Selker, E.U. 1990. DNA methylation and chromatin structure: states induced by the Polycomb and trithorax group genes. A view from below. Trends Biochem. Sci. 15: 103–107. Novartis Found. Symp. 214: 51–61. ———. 1999. Gene silencing: Repeats that count. Cell 97: 157– Pfeifer, G.P., Steigerwald, S.D., Hansen, R.S., Gartler, S.M., and 160. Riggs, A.D. 1990. Polymerase chain reaction-aided genomic Shemer, R., Hershko, A.Y., Perk, J., Mostoslavsky, R., Tsuberi, sequencing of an X chromosome-linked CpG island: Meth- B., Cedar, H., Buiting, K., and Razin, A. 2000. The imprinting ylation patterns suggest clonal inheritance, CpG site au- box of the Prader-Willi/Angelman syndrome domain. Nat. tonomy, and an explanation of activity state stability. Proc. Genet. 26: 440–443. Nat. Acad. Sci. 87: 8252–8256. Siegfried, Z., Eden, S., Mendelsohn, M., Feng, X., Tsuberi, B.Z., Pirrotta, V. 1999. Polycomb silencing and the maintenance of and Cedar, H. 1999. DNA methylation represses transcrip- stable chromatin states. Results Probl. Cell. Differ. 25: 205– tion in vivo. Nat. Genet. 22: 203–206. 228. Silva, A.J., Ward, K., and White, R. 1993. Mosaic methylation in Pollack, Y., Stein, R., Razin, A., and Cedar, H. 1980. Methyl- clonal tissue. Dev. Biol. 156: 391–398. ation of foreign DNA sequences in eukaryotic cells. Proc. Simmen, M.W., Leitgeb, S., Charlton, J., Jones, S.J.M., Harris, Natl. Acad. Sci. 77: 6463–6467. B.R., Clark, V.H., and Bird, A. 1999. Nonmethylated trans- Poux, S., McCabe, D., and Pirrotta, V. 2001. Recruitment of posable elements and methylated genes in a chordate ge- components of Polycomb Group chromatin complexes in nome. Science 283: 1164–1167. Drosophila. Development 128: 75–85. Smit, A.F. 1999. Interspersed repeats and other mementos of Pradhan, S., Bacolla, A., Wells, R.D., and Roberts, R.J. 1999. transposable elements in mammalian genomes. Curr. Opin. Recombinant human DNA (cytosine-5) methyltransferase. I. Genet. Dev. 9: 657–663. Expression, purification, and comparison of de novo and Smith, S.S., Kan, J.L.C., Baker, D.J., Kaplan, B.E., and Dembek, maintenance methylation. J. Biol. Chem. 274: 33002–33010. P. 1991. Recognition of unusual DNA structures by human Prokhortchouk, A., Hendrich, B., Jorgensen, H., Ruzov, A., DNA(cytosine-5)methyltransferase. J. Mol. Biol. 217: 39–51. Wilm, M., Georgiev, G., Bird, A., and Prokhortchouk, E. Stancheva, I. and Meehan, R.R. 2000. Transient depletion of 2001. The p120 catenin partner Kaiso is a DNA methylation- xDnmt1 leads to premature gene activation in Xenopus em- dependent transcriptional repressor. Genes & Dev. 15: 1613– bryos. Genes & Dev. 14: 313–327. 1618. Stewart, C.L., Stuhlmann, H., Jahner, D., and Jaenisch, R. 1982. Ramchandani, S., Bhattacharya, S.K., Cervoni, N., and Szyf, M. De novo methylation and infectivity of retroviral genomes 1999. DNA methylation is a reversible biological signal. introduced into embryonal carcinoma cells. Proc. Natl. Proc. Natl. Acad. Sci. 96: 6107–6112. Acad. Sci. 79: 4098–4102. Ramsahoye, B.H., Biniszkiewicz, D., Lyko, F., Clark, V., Bird, Stöger, R., Kajimura, T.M., Brown, W.T., and Laird, C.D. 1997. A.P., and Jaenisch, R. 2000. Non-CpG methylation is preva- Epigenetic variation illustrated by DNA methylation pat- lent in embryonic stem cells and may be mediated by DNA terns of the fragile-X gene FMR1. Hum. Mol. Gen. 6: 1791– methyltransferase 3a. Proc. Natl. Acad. Sci. 97: 5237–5242. 1801. Reik, W. and Walter, J. 2001. Evolution of imprinting mecha- Swisher, J.F., Rand, E., Cedar, H., and Marie Pyle, A. 1998. nisms: The battle of the sexes begins in the zygote. Nat. Analysis of putative RNase sensitivity and protease insensi- Genet. 27: 255–256. tivity of demethylation activity in extracts from rat myo- Reik, W., Dean, W., and Walter, J. 2001. Epigenetic reprogram- blasts. Nucleic Acids Res. 26: 5573–5580. ming in mammalian development. Science 293: 1089–1093. Szabo, P., Tang, S.H., Rentsendorj, A., Pfeifer, G.P., and Mann, Rhee, I., Jair, K.W., Yen, R.W., Lengauer, C., Herman, J.G., Kin- J.R. 2000. Maternal-specific footprints at putative CTCF zler, K.W., Vogelstein, B., Baylin, S.B., and Schuebel, K.E. sites in the H19 imprinting control region give evidence for 2000. CpG methylation is maintained in human cancer cells insulator function. Curr. Biol. 10: 607–610. lacking DNMT1. Nature 404: 1003–1007. Tada, M., Tada, T., Lefebvre, L., Barton, S.C., and Surani, M.A. Rideout, I.W., Eggan, K., and Jaenisch, R. 2001. Nuclear cloning 1997. Embryonic germ cells induce epigenetic reprogram-

20 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

DNA methylation and epigenetic memory

ming of somatic nucleus in hybrid cells. EMBO J. 16: 6510– Weng, A., Magnuson, T., and Storb, U. 1995. Strain-specific 6520. transgene methylation occurs early in mouse development Tamaru, H. and Selker, E.U. 2001. A histone H3 methyltrans- and can be recapitulated in embryonic stem cells. Develop- ferase controls DNA methylation in Neurospora crassa. ment 121: 2853–2859. Nature 414: 277–283. Wigler, M., Levy, D., and Perucho, M. 1981. The somatic repli- Tazi, J. and Bird, A. 1990. Alternative chromatin structure at cation of DNA methylation. Cell 24: 33–40. CpG islands. Cell 60: 909–920. Wolffe, A.P., Jones, P.L., and Wade, P.A. 1999. DNA demethyl- Thomassin, H., Flavin, M., Espinas, M.L., and Grange, T. 2001. ation. Proc. Natl. Acad. Sci. 96: 5894–5896. Glucocorticoid-induced DNA demethylation and gene Woodcock, D.M., Lawler, C.B., Linsenmeyer, M.E., Doherty, memory during development. EMBO J. 20: 1974–1983. J.P., and Warren, W.D. 1997. Asymmetric methylation in the Tremblay, K.D., Duran, K.L., and Bartolomei, M.S. 1997. A 5Ј hypermethylated CpG promoter region of the human L1 ret- 2-kilobase-pair region of the imprinted mouse H19 gene ex- rotransposon. J. Biol. Chem 272: 7810–7816. hibits exclusive paternal methylation throughout develop- Wutz, A. and Jaenisch, R. 2000. A shift from reversible to irre- ment. Mol. Cell. Biol. 17: 4322–4329. versible X inactivation is triggered during ES cell differen- Turker, M.S. 1999. The establishment and maintenance of tiation. Mol. Cell 5: 695–705. DNA methylation patterns in mouse somatic cells. Semin. Wutz, A., Smrzka, O.W., Schweifer, N., Schellander, K., Wag- Cancer Biol. 9: 329–337. ner, E.F., and Barlow, D.P. 1997. Imprinted expression of the Tweedie, S., Charlton, J., Clark, V., and Bird, A. 1997. Methyl- Igf2r gene depends on an intronic CpG island. Nature ation of genomes and genes at the invertebrate–vertebrate 389: 745–749. boundary. Mol. Cell. Biol. 17: 1469–1475. Xu, G.-L., Bestor, T.H., Bourc’his, D., Hsieh, C.-L., Tommerup, Tweedie, S., Ng, H.-H., Barlow, A.L., Turner, B.M., Hendrich, B., N., Bugge, M., Hulten, M., Qu, X., Russo, J.J., and Viegas- and Bird, A. 1999. Vestiges of a DNA methylation system in Pequignot, E. 1999. Chromosome instability and immuno- Drosophila melanogaster. Nat. Genet. 23: 389–390. deficiency syndrome caused by mutations in a DNA meth- Um, S., Harbers, M., Benecke, A., Pierrat, B., Losson, R., and yltransferase gene. Nature 402: 187–191. Chambon, P. 1998. Retinoic acid receptors interact physi- Yates, P.A., Burman, R.W., Mummaneni, P., Krussel, S., and cally and functionally with the T:G mismatch-specific thy- Turker, M.S. 1999. Tandem B1 elements located in a mouse mine-DNA glycosylase. J. Biol. Chem. 273: 20728–20736. methylation center provide a target for de novo DNA meth- Venolia, L., Gartler, S.M., Wasserman, E.R., Yen, P., Mohandas, ylation. J. Biol. Chem. 274: 36357–36361. T., and Shapiro, L.J. 1982. Transformation with DNA from 5 Yoder, J.A., Walsh, C.P., and Bestor, T.H. 1997. Cytosine meth- azacytidine-reactivated X chromosomes. Proc. Natl. Acad. ylation and the ecology of intragenomic parasites. Trends Sci. 79: 2352–2354. Genet. 13: 335–340. Venter, J.C., Adams, M.D., Myers, E.W., Li, P.W., Mural, R.J., Zhang, Y. and Reinberg, D. 2001. Transcription regulation by Sutton, G.G., Smith, H.O., Yandell, M., Evans, C.A., Holt, histone methylation: Interplay between different covalent R.A., et al. 2001. The sequence of the human genome. Sci- modifications of the core histone tails. Genes & Dev. ence 291: 1304–1351. 15: 2343–2360. Viegas-Pequignot, E. and Dutrillaux, B. 1976. Segmentation of Zhu, B., Zheng, Y., Angliker, H., Schwarz, S., Thiry, S., Sieg- human chromosomes induced by 5-ACR (5-azacytidine). mann, M., and Jost, J.P. 2000a. 5-Methylcytosine DNA gly- Hum. Genet. 34: 247–254. cosylase activity is also present in the human MBD4 (G/T Voo, K.S., Carlone, D.L., Jacobsen, B.M., Flodin, A., and Skalnik, mismatch glycosylase) and in a related avian sequence. D.G. 2000. Cloning of a mammalian transcriptional activa- Nucleic Acids Res. 28: 4157–4165. tor that binds unmethylated CpG motifs and shares a CXXC Zhu, B., Zheng, Y., Hess, D., Angliker, H., Schwarz, S., Sieg- domain with DNA methyltransferase, human trithorax, and mann, M., Thiry, S., and Jost, J.-P. 2000b. 5-Methylcytosine- methyl-CpG binding domain protein 1. Mol. Cell. Biol. DNA glycosylase activity is present in a cloned G/T mis- 20: 2108–2121. match DNA glycosylase associated with the chicken em- Wakayama, T. and Yanagimachi, R. 2001. Mouse cloning with bryo DNA demethylation complex. Proc. Natl. Acad. Sci. nucleus donor cells of different age and type. Mol. Reprod. 97: 5135–5139. Dev. 58: 376–383. Zhu, B., Benjamin, D., Zheng, Y., Angliker, H., Thiry, S., Sieg- Walsh, C.P. and Bestor, T.H. 1999. Cytosine methylation and mann, M., and Jost, J.P. 2001. Overexpression of 5-methyl- mammalian development. Genes & Dev. 13: 26–34. cytosine DNA glycosylase in human embryonic kidney cells Walsh, C.P., Chaillet, J.R., and Bestor, T.H. 1998. Transcription EcR293 demethylates the promoter of a hormone-regulated of IAP endogenous retroviruses is constrained by cytosine reporter gene. Proc. Natl. Acad. Sci. 98: 5031–5036. methylation. Nat. Genet. 20: 116–117. Wang, J., Mager, J., Chen, Y., Schneider, E., Cross, J.C., Nagy, A., and Magnuson, T. 2001. Imprinted X inactivation main- tained by a mouse Polycomb group gene. Nat. Genet. 28: 371–375. Wassenegger, M., Heimes, S., Riedel, L., and Sanger, H.L. 1994. RNA-directed de novo methylation of genomic sequences in plants. Cell 76: 567–576. Waugh O’Neill, R.J., O’Neill, M.J., and Marshall Graves, J.A. 1998. Undermethylation associated with retroelement acti- vation and chromosome remodelling in an interspecific mammalian hybrid. Nature 393: 68–72. Weiss, A., Keshet, I., Razin, A., and Cedar, H. 1996. DNA de- methylation in vitro: Involvement of RNA. Cell 86: 709– 718.

GENES & DEVELOPMENT 21 Downloaded from genesdev.cshlp.org on September 24, 2021 - Published by Cold Spring Harbor Laboratory Press

DNA methylation patterns and epigenetic memory

Adrian Bird

Genes Dev. 2002, 16: Access the most recent version at doi:10.1101/gad.947102

References This article cites 197 articles, 70 of which can be accessed free at: http://genesdev.cshlp.org/content/16/1/6.full.html#ref-list-1

License

Email Alerting Receive free email alerts when new articles cite this article - sign up in the box at the top Service right corner of the article or click here.

Cold Spring Harbor Laboratory Press